Issues with ocn2glc_coupling

[mvdebolskiy]

Right now, there is no check for this and setting ocn2glc_coupling = .true. in user_nl_cpl will just give you an error during runtime. F.e.

(med_phases_prep_glc_init) mapbilnr is not created for ocn->glc mapping

Which is not verbose enough.
Instead, buildnml should explicitly say that ocn2glc_coupling can only be true if running with DOCN%MULTILEV or BLOM

[mvdebolskiy]

Looking through the routine calls and runseq:
med_phases_prep_glc_avg is called in two places:

1. In med_phases_prep_glc when lnd2glc_coupling is true.
2. In med_phases_post_lnd when histaux_l2x1yrg is present and set and true.

Now, let's look at med_phases_accum_ocn and med_phases_accum_lnd. The latter is only called in post land and in case 2, averaging is called immediately. med_phases_accum_ocn On the other hand is only called in med_phases_post_ocn

We use histaux_l2x1yrg together with ocn2glc_coupling to write out coupler history fields from ocn to glc. However, in MOST of the run sequences med_phases_post_lnd is run before med_phases_post_ocn. Meaning, that the ocean fields are not correct averages that are getting written in in the auxillary files.

@mvertens

[mvdebolskiy]

ERS tests fails with TC* compsets (lnd and ocn data to glc).
@mvertens and @mpetrini-norce found that on restart glc1Exp_So* fields differ between continuous run and a restart run. The differences appear only on the next year after the restart. For all other years the fields are the same. Something during the first timestep in the mediator makes the fields differ on restart. Fields are different by a tiny bit everywhere, not just at 1-2 gridcells.

[mvertens]

@mpetrini-norce @mvdebolskiy @matsbn - I think that ocn2glc_coupling is not being set correctly. The current logic in the cmeps buildnml is

     config["ocn_multilev"] = (
         ".true." if case.get_value("COMP_OCN") == 'blom' or 'DOCN%MULTILEV' in case.get_value("COMPSET") else ".false."

So that means that if you are running with BLOM, you will be sending all the multi level fields to the mediator even if CISM is not active. I think this is incorrect. I propose that we change this to

     config["ocn_multilev"] = (
        ".true." if case.get_value("COMP_GLC" == 'cism') and
        (case.get_value("COMP_OCN") == 'blom' or 'DOCN%MULTILEV' in case.get_value("COMPSET")) else ".false."

I just want to make sure that I am not missing anything in the logic.
In addituibu - are ERS tests failing with TC compsets still?

[matsbn]

@mvertens, the logic you propose for setting ocn_multilev seems fine to me.

[mvdebolskiy]

@mvertens the restart issues are fixed, however, the question of ocean data is behind by one timestep is still open.

[mvdebolskiy]

@mvertens sending ocn fields to glc is fine, since we need histaux_l2x1yrg together with ocn2glc_coupling to be true to write out ocn coupler history to glc even if cism is not on.